|
17 | 17 | */ |
18 | 18 | package org.apache.beam.sdk.schemas; |
19 | 19 |
|
| 20 | +import java.util.Arrays; |
| 21 | +import java.util.List; |
| 22 | +import java.util.Map; |
| 23 | +import java.util.Objects; |
20 | 24 | import org.apache.beam.sdk.schemas.Schema.FieldType; |
21 | 25 | import org.apache.beam.sdk.schemas.Schema.LogicalType; |
| 26 | +import org.apache.beam.sdk.values.Row; |
22 | 27 |
|
23 | 28 | /** A set of utility functions for schemas. */ |
24 | 29 | @SuppressWarnings({ |
25 | 30 | "nullness" // TODO(https://github.com/apache/beam/issues/20497) |
26 | 31 | }) |
27 | 32 | public class SchemaUtils { |
| 33 | + private static final String INDENT = " "; |
| 34 | + |
28 | 35 | /** |
29 | 36 | * Given two schema that have matching types, return a nullable-widened schema. |
30 | 37 | * |
@@ -122,4 +129,276 @@ public static <BaseT, InputT> InputT toLogicalInputType( |
122 | 129 | LogicalType<InputT, BaseT> logicalType, BaseT baseType) { |
123 | 130 | return logicalType.toInputType(baseType); |
124 | 131 | } |
| 132 | + |
| 133 | + public static String toPrettyString(Row row) { |
| 134 | + return toPrettyRowString(row, ""); |
| 135 | + } |
| 136 | + |
| 137 | + public static String toPrettyString(Schema schema) { |
| 138 | + return toPrettySchemaString(schema, ""); |
| 139 | + } |
| 140 | + |
| 141 | + static String toFieldTypeNameString(FieldType fieldType) { |
| 142 | + return fieldType.getTypeName() |
| 143 | + + (Boolean.TRUE.equals(fieldType.getNullable()) ? "" : " NOT NULL"); |
| 144 | + } |
| 145 | + |
| 146 | + static String toPrettyFieldTypeString(Schema.FieldType fieldType, String prefix) { |
| 147 | + String nextPrefix = prefix + INDENT; |
| 148 | + switch (fieldType.getTypeName()) { |
| 149 | + case BYTE: |
| 150 | + case INT16: |
| 151 | + case INT32: |
| 152 | + case INT64: |
| 153 | + case DECIMAL: |
| 154 | + case FLOAT: |
| 155 | + case DOUBLE: |
| 156 | + case STRING: |
| 157 | + case DATETIME: |
| 158 | + case BOOLEAN: |
| 159 | + case BYTES: |
| 160 | + return "<" + toFieldTypeNameString(fieldType) + ">"; |
| 161 | + case ARRAY: |
| 162 | + case ITERABLE: |
| 163 | + { |
| 164 | + StringBuilder sb = new StringBuilder(); |
| 165 | + sb.append("<").append(toFieldTypeNameString(fieldType)).append("> {\n"); |
| 166 | + sb.append(nextPrefix) |
| 167 | + .append("<element>: ") |
| 168 | + .append( |
| 169 | + toPrettyFieldTypeString( |
| 170 | + Objects.requireNonNull(fieldType.getCollectionElementType()), nextPrefix)) |
| 171 | + .append("\n"); |
| 172 | + sb.append(prefix).append("}"); |
| 173 | + return sb.toString(); |
| 174 | + } |
| 175 | + case MAP: |
| 176 | + { |
| 177 | + StringBuilder sb = new StringBuilder(); |
| 178 | + sb.append("<").append(toFieldTypeNameString(fieldType)).append("> {\n"); |
| 179 | + sb.append(nextPrefix) |
| 180 | + .append("<key>: ") |
| 181 | + .append( |
| 182 | + toPrettyFieldTypeString( |
| 183 | + Objects.requireNonNull(fieldType.getMapKeyType()), nextPrefix)) |
| 184 | + .append(",\n"); |
| 185 | + sb.append(nextPrefix) |
| 186 | + .append("<value>: ") |
| 187 | + .append( |
| 188 | + toPrettyFieldTypeString( |
| 189 | + Objects.requireNonNull(fieldType.getMapValueType()), nextPrefix)) |
| 190 | + .append("\n"); |
| 191 | + sb.append(prefix).append("}"); |
| 192 | + return sb.toString(); |
| 193 | + } |
| 194 | + case ROW: |
| 195 | + { |
| 196 | + return "<" |
| 197 | + + toFieldTypeNameString(fieldType) |
| 198 | + + "> " |
| 199 | + + toPrettySchemaString(Objects.requireNonNull(fieldType.getRowSchema()), prefix); |
| 200 | + } |
| 201 | + case LOGICAL_TYPE: |
| 202 | + { |
| 203 | + Schema.FieldType baseType = |
| 204 | + Objects.requireNonNull(fieldType.getLogicalType()).getBaseType(); |
| 205 | + StringBuilder sb = new StringBuilder(); |
| 206 | + sb.append("<") |
| 207 | + .append(toFieldTypeNameString(fieldType)) |
| 208 | + .append("(") |
| 209 | + .append(fieldType.getLogicalType().getIdentifier()) |
| 210 | + .append(")> {\n"); |
| 211 | + sb.append(nextPrefix) |
| 212 | + .append("<base>: ") |
| 213 | + .append(toPrettyFieldTypeString(baseType, nextPrefix)) |
| 214 | + .append("\n"); |
| 215 | + sb.append(prefix).append("}"); |
| 216 | + return sb.toString(); |
| 217 | + } |
| 218 | + default: |
| 219 | + throw new UnsupportedOperationException(fieldType.getTypeName() + " is not supported"); |
| 220 | + } |
| 221 | + } |
| 222 | + |
| 223 | + static String toPrettyOptionsString(Schema.Options options, String prefix) { |
| 224 | + String nextPrefix = prefix + INDENT; |
| 225 | + StringBuilder sb = new StringBuilder(); |
| 226 | + sb.append("{\n"); |
| 227 | + for (String optionName : options.getOptionNames()) { |
| 228 | + sb.append(nextPrefix) |
| 229 | + .append(optionName) |
| 230 | + .append(" = ") |
| 231 | + .append( |
| 232 | + toPrettyFieldValueString( |
| 233 | + options.getType(optionName), options.getValue(optionName), nextPrefix)) |
| 234 | + .append("\n"); |
| 235 | + } |
| 236 | + sb.append(prefix).append("}"); |
| 237 | + return sb.toString(); |
| 238 | + } |
| 239 | + |
| 240 | + static String toPrettyFieldValueString(Schema.FieldType fieldType, Object value, String prefix) { |
| 241 | + String nextPrefix = prefix + INDENT; |
| 242 | + switch (fieldType.getTypeName()) { |
| 243 | + case BYTE: |
| 244 | + case INT16: |
| 245 | + case INT32: |
| 246 | + case INT64: |
| 247 | + case DECIMAL: |
| 248 | + case FLOAT: |
| 249 | + case DOUBLE: |
| 250 | + case DATETIME: |
| 251 | + case BOOLEAN: |
| 252 | + return Objects.toString(value); |
| 253 | + case STRING: |
| 254 | + { |
| 255 | + String string = (String) value; |
| 256 | + return "\"" + string.replace("\\", "\\\\").replace("\"", "\\\"") + "\""; |
| 257 | + } |
| 258 | + case BYTES: |
| 259 | + { |
| 260 | + byte[] bytes = (byte[]) value; |
| 261 | + return Arrays.toString(bytes); |
| 262 | + } |
| 263 | + case ARRAY: |
| 264 | + case ITERABLE: |
| 265 | + { |
| 266 | + if (!(value instanceof List)) { |
| 267 | + throw new IllegalArgumentException( |
| 268 | + String.format( |
| 269 | + "value type is '%s' for field type '%s'", |
| 270 | + value.getClass(), fieldType.getTypeName())); |
| 271 | + } |
| 272 | + FieldType elementType = Objects.requireNonNull(fieldType.getCollectionElementType()); |
| 273 | + |
| 274 | + @SuppressWarnings("unchecked") |
| 275 | + List<Object> list = (List<Object>) value; |
| 276 | + if (list.isEmpty()) { |
| 277 | + return "[]"; |
| 278 | + } |
| 279 | + StringBuilder sb = new StringBuilder(); |
| 280 | + sb.append("[\n"); |
| 281 | + int size = list.size(); |
| 282 | + int index = 0; |
| 283 | + for (Object element : list) { |
| 284 | + sb.append(nextPrefix) |
| 285 | + .append(toPrettyFieldValueString(elementType, element, nextPrefix)); |
| 286 | + if (index++ < size - 1) { |
| 287 | + sb.append(",\n"); |
| 288 | + } else { |
| 289 | + sb.append("\n"); |
| 290 | + } |
| 291 | + } |
| 292 | + sb.append(prefix).append("]"); |
| 293 | + return sb.toString(); |
| 294 | + } |
| 295 | + case MAP: |
| 296 | + { |
| 297 | + if (!(value instanceof Map)) { |
| 298 | + throw new IllegalArgumentException( |
| 299 | + String.format( |
| 300 | + "value type is '%s' for field type '%s'", |
| 301 | + value.getClass(), fieldType.getTypeName())); |
| 302 | + } |
| 303 | + |
| 304 | + FieldType keyType = Objects.requireNonNull(fieldType.getMapKeyType()); |
| 305 | + FieldType valueType = Objects.requireNonNull(fieldType.getMapValueType()); |
| 306 | + |
| 307 | + @SuppressWarnings("unchecked") |
| 308 | + Map<Object, Object> map = (Map<Object, Object>) value; |
| 309 | + if (map.isEmpty()) { |
| 310 | + return "{}"; |
| 311 | + } |
| 312 | + |
| 313 | + StringBuilder sb = new StringBuilder(); |
| 314 | + sb.append("{\n"); |
| 315 | + int size = map.size(); |
| 316 | + int index = 0; |
| 317 | + for (Map.Entry<Object, Object> entry : map.entrySet()) { |
| 318 | + sb.append(nextPrefix) |
| 319 | + .append(toPrettyFieldValueString(keyType, entry.getKey(), nextPrefix)) |
| 320 | + .append(": ") |
| 321 | + .append(toPrettyFieldValueString(valueType, entry.getValue(), nextPrefix)); |
| 322 | + if (index++ < size - 1) { |
| 323 | + sb.append(",\n"); |
| 324 | + } else { |
| 325 | + sb.append("\n"); |
| 326 | + } |
| 327 | + } |
| 328 | + sb.append(prefix).append("}"); |
| 329 | + return sb.toString(); |
| 330 | + } |
| 331 | + case ROW: |
| 332 | + { |
| 333 | + return toPrettyRowString((Row) value, prefix); |
| 334 | + } |
| 335 | + case LOGICAL_TYPE: |
| 336 | + { |
| 337 | + @SuppressWarnings("unchecked") |
| 338 | + Schema.LogicalType<Object, Object> logicalType = |
| 339 | + (Schema.LogicalType<Object, Object>) |
| 340 | + Objects.requireNonNull(fieldType.getLogicalType()); |
| 341 | + Schema.FieldType baseType = logicalType.getBaseType(); |
| 342 | + Object baseValue = logicalType.toBaseType(value); |
| 343 | + return toPrettyFieldValueString(baseType, baseValue, prefix); |
| 344 | + } |
| 345 | + default: |
| 346 | + throw new UnsupportedOperationException(fieldType.getTypeName() + " is not supported"); |
| 347 | + } |
| 348 | + } |
| 349 | + |
| 350 | + static String toPrettySchemaString(Schema schema, String prefix) { |
| 351 | + String nextPrefix = prefix + INDENT; |
| 352 | + StringBuilder sb = new StringBuilder(); |
| 353 | + sb.append("{\n"); |
| 354 | + for (Schema.Field field : schema.getFields()) { |
| 355 | + sb.append(nextPrefix) |
| 356 | + .append(field.getName()) |
| 357 | + .append(": ") |
| 358 | + .append(toPrettyFieldTypeString(field.getType(), nextPrefix)); |
| 359 | + if (field.getOptions().hasOptions()) { |
| 360 | + sb.append(", fieldOptions = ") |
| 361 | + .append(toPrettyOptionsString(field.getOptions(), nextPrefix)); |
| 362 | + } |
| 363 | + sb.append("\n"); |
| 364 | + } |
| 365 | + sb.append(prefix).append("}"); |
| 366 | + if (schema.getOptions().hasOptions()) { |
| 367 | + sb.append(", schemaOptions = ").append(toPrettyOptionsString(schema.getOptions(), prefix)); |
| 368 | + } |
| 369 | + if (schema.getUUID() != null) { |
| 370 | + sb.append(", schemaUUID = ").append(schema.getUUID()); |
| 371 | + } |
| 372 | + return sb.toString(); |
| 373 | + } |
| 374 | + |
| 375 | + static String toPrettyRowString(Row row, String prefix) { |
| 376 | + long nonNullFieldCount = row.getValues().stream().filter(Objects::nonNull).count(); |
| 377 | + if (nonNullFieldCount == 0) { |
| 378 | + return "{}"; |
| 379 | + } |
| 380 | + |
| 381 | + String nextPrefix = prefix + INDENT; |
| 382 | + StringBuilder sb = new StringBuilder(); |
| 383 | + sb.append("{\n"); |
| 384 | + long nonNullFieldIndex = 0; |
| 385 | + for (Schema.Field field : row.getSchema().getFields()) { |
| 386 | + String fieldName = field.getName(); |
| 387 | + Object fieldValue = row.getValue(fieldName); |
| 388 | + if (fieldValue == null) { |
| 389 | + continue; |
| 390 | + } |
| 391 | + sb.append(nextPrefix) |
| 392 | + .append(fieldName) |
| 393 | + .append(": ") |
| 394 | + .append(toPrettyFieldValueString(field.getType(), fieldValue, nextPrefix)); |
| 395 | + if (nonNullFieldIndex++ < nonNullFieldCount - 1) { |
| 396 | + sb.append(",\n"); |
| 397 | + } else { |
| 398 | + sb.append("\n"); |
| 399 | + } |
| 400 | + } |
| 401 | + sb.append(prefix).append("}"); |
| 402 | + return sb.toString(); |
| 403 | + } |
125 | 404 | } |
0 commit comments